BTLA identifies dysfunctional PD-1-expressing CD4⁺ T cells in human hepatocellular carcinoma

ABSTRACT

Although immunotherapy targeting programmed cell death 1/programmed cell death ligand 1 (PD-1/PD-L1) pathway is being applied in clinic, the response outcomes are heterogeneous, suggesting existences of distinctive subsets within PD-1-expressing T cells that react differently to PD-1/PD-L1 blockade. However, markers to demarcate these subsets in human cancers remain unclear. Here, we found that both PD-1 and B and T lymphocyte attenuator (BTLA) were significantly upregulated on CD4⁺ T cells from tumor compared with those from paired non-tumor liver in hepatocellular carcinoma (HCC) patients. Interestingly, over 85% BTLA⁺ CD4⁺ T cells were PD-1-expressing cells and represented about 50% PD-1⁺ CD4⁺ T cells in tumors, and that level of BTLA⁺PD-1⁺ tumor CD4⁺ T cells were selectively associated with advanced stage HCC. BTLA⁺ identified highly dysfunctional PD-1-expressing CD4⁺ T cell subset, whereas BTLA⁻ defined PD-1⁺ CD4⁺ T cells undergoing activation in HCC. Importantly, blockade of PD-L1 could restore the ability of IFNγ/TNF-α production in BTLA⁺PD-1⁺ tumor CD4⁺ T cells but partially suppressed the activation of BTLA⁻PD-1⁺ CD4⁺ T cells. Moreover, we provided evidence that BTLA signals also participated in suppressing CD4⁺ T cell function in HCC. In conclusion, BTLA could identify distinct function of PD-1 expressing CD4⁺ T cells in human cancer, which might not only advance our understanding of inhibitory receptor blockade, but also provide new targets for clinical predictors of response to these immunotherapies.

Abbreviations

BTLA

B and T lymphocyte attenuator

FACS

fluorescence-activated cell sorting

HCC

hepatocellular carcinoma

HVEM

herpesvirus entry mediator

mAb

monoclonal antibody

MACS

magnetic cell separation

PD-1

programmed cell death 1

xTNM

tumor node metastases

Introduction

The elevated expression of programmed cell death 1 (PD-1) on tumor-infiltrating T cells has become an accepted signature for a state called T cell “exhaustion” in cancer.^1-4 The blockade of PD-1 can reverse T cell exhaustion and restore antigen-specific T cell responses, highlighting the therapeutic significance of PD-1.^5-9 However, immunotherapy strategies targeting the PD-1/PD-L1 pathway fails to result in a complete restoration of T cell function in human hepatocellular carcinoma (HCC) and some other cancers.^5,10,11 More surprisingly, emerging evidence shows that PD-1-expressing tumor-infiltrating T cells are in fact a favorable prognostic biomarker in some cancers.^12-14 These controversies in literatures suggest that PD-1 is not simply a marker of “exhaustion,’ but can serve as an indicator of ongoing immune response.^1,12,15,16 There is a pressing need to introduce additional biomarker(s) to sub-classify the PD-1-expressing T cell subsets and to better direct clinical application of PD-1-based immunotherapy in human cancer.^17,18

Beyond the PD-1 pathway, T cell inhibition mediated by the BTLA pathway (B and T lymphocyte attenuator, CD272) is now one of the most studied.^19-21 In human melanoma, persistent expression of BTLA has been reported on tumor antigen-specific effector CD8⁺ T cells.^22-24 Studies have shown that tumor-specific CD8⁺ T cells persistently expressed high levels of BTLA in vivo and remained susceptible to functional inhibition by its ligand herpesvirus entry mediator (HVEM).²³ Much less in known whether BTLA and PD-1 are expressed on distinct or overlapping populations of CD4⁺ T cells in human tumors.^22,24 In the current study, we showed that over 85% BTLA⁺ CD4⁺ T cells were PD-1-expressing cells and represented about 50% of the PD-1⁺ CD4⁺ T cells in human HCC. BTLA⁺ identified a highly dysfunctional PD-1-expressing CD4⁺ T cell subset, whereas BTLA⁻ defined PD-1⁺ CD4⁺ T cells that were undergoing activation in HCC. Importantly, blockade of PD-L1 restored the IFNγ/TNF-α production in BTLA⁺PD-1⁺ tumor CD4⁺ T cells but partially suppressed the activation of BTLA⁻PD-1⁺ CD4⁺ T cells. Moreover, we provided evidence that BTLA signals also participated in suppressing CD4⁺ T cell function in tumor. Therefore, BTLA could identify distinct function of PD-1-expressing CD4⁺ T cells in human cancer, which might have important implications for our understanding of inhibitory receptor blockade and developing better cancer immunotherapeutic strategies.

Results

BTLA⁺ identifies a PD-1-expressing CD4⁺ T cell subset that correlates with advanced stage HCC

We first used flow cytometry to analyze the expression of co-inhibitory receptor BTLA and PD-1 on CD4⁺ T cells from HCC tumor tissues paired with non-tumor liver tissues (Table 1). In the samples analyzed, we observed a significantly increased percentage of BTLA⁺ CD4⁺ T cells with higher BTLA intensity in tumor tissue: at least 2-fold increase of BTLA⁺ CD4⁺ T cell proportion was detected in tumor as compared with paired non-tumor liver (48 ± 5.7% vs. 24 ± 4.2%, n = 19, p = 0.0027; Figs. 1A and B). Although the proportion of PD-1⁺ CD4⁺ T cells was significantly higher than that of BTLA⁺ CD4⁺ T cells in tumor, only a 1.5-fold increase in PD-1⁺ CD4⁺ T cell proportion was detected in tumor as compared with paired non-tumor liver (63 ± 4.1% vs. 42 ± 3.1%, n = 36, p < 0.0001; Figs. 1A and B).

Table 1.

Clinical characteristics of the 49 HCC patients.

Patient characteristics
No. of patients	49
Age, years (median, range)	48, 22–69
Gender (male/female)	45/4
HbsAg (negative/positive)	5/44
Cirrhosis (absent/present)	14/35
ALT, U/L (median, range)	43, 16–736
AFP, ng/mL (≤25/>25)	17/32
Tumor size, cm (≤5/>5)	10/39
Tumor multiplicity (solitary/multiple)	32/17
Vascular invasion (absent/present)	35/14
Intrahepatic metastasis (no/yes)	38/11
TNM stage (I+II/III+IV)	26/23
Tumor differentiation (I+II/III+IV)	31/18
Fibrous capsule (absent/present)	21/28

Abbreviations: HbsAg, hepatitis B surface antigen; ALT, alanine aminotransferase; AFP, α-fetoprotein; TNM, tumor node metastasis; ND, not determined.

Figure 1.

Co-expression of BTLA and PD-1 identifies a CD4⁺ T cell subset that correlates with disease progression in HCC patients. (A and B) Increased expression of BTLA (n = 19) and PD-1 (n = 36) on tumor-infiltrating CD4⁺ T cells from HCC patients. Horizontal bars represent mean values (B). (C and D) Co-expression patterns of BTLA and PD-1 on non-tumor and tumor-infiltrating CD4⁺ T cells (n = 14). (E) The percentage of BTLA⁺PD-1⁺ CD4⁺ T cells (upper) and BTLA⁻PD-1⁺ CD4⁺ T cells (lower) according to TNM stage (stage I and II [n = 5] vs. stages III and IV [n = 9]). Horizontal bars represent median values. The p-values were calculated using the Student's t test.

Inasmuch as BTLA and PD-1 were both upregulated in tumor CD4⁺ T cells, we then asked whether BTLA and PD-1-identified distinct T helper cell subsets in human HCC. Strikingly, 83 ± 6.5% of the BTLA-expressing tumor CD4⁺ T cells were PD-1⁺, whereas only 54 ± 7.9% of the PD-1-expressing tumor CD4⁺ T cells were BTLA⁺ (n = 14, Figs. 1C and D), suggesting that BTLA is likely to identify distinct PD-1-expressing CD4⁺ T cell subsets in tumors. Consistent with this, we found that the frequency of tumor BTLA⁺PD-1⁺ CD4⁺ T cells was significantly increased in patients with advanced stage HCC (stage I and II [n = 5] vs. stages III and IV [n = 9]; p = 0.0097; Fig. 1E). However, increased proportion of tumor BTLA⁻PD-1⁺ CD4⁺ T cells was not associated with disease progression of HCC (Fig. 1E).

BTLA⁺PD-1⁺, but not BTLA⁻PD-1⁺, CD4⁺ T cells exhibit an exhausted phenotype

Having established that BTLA⁺PD-1⁺ CD4⁺ T cell infiltration positively correlated with advanced stage HCC, we then compared the phenotypic and functional features of BTLA⁺ and PD-1⁺ CD4⁺ T cells. We examined the ability of freshly isolated tumor/non-tumor-infiltrating T cells to produce cytokines ex vivo, upon short stimulation with Leukocyte Activation Cocktail as describe in Methods. In general, the BTLA⁺ CD4⁺ T cells derived from both tumor and paired non-tumor liver exhibited reduced capacity to produce IFNγ (p < 0.01 compared with BTLA⁻ CD4⁺ T cells, n = 13, Figs. 2A and B). Interestingly, the PD-1⁺ CD4⁺ T cells derived from tumor and paired non-tumor liver displayed opposed IFNγ production profile: although most PD-1⁺ CD4⁺ T cells derived from tumor tissue had become exhausted to produce IFNγ, their counterparts in non-tumor liver did have more potential to produce IFNγ compared with PD-1⁻ CD4⁺ T cells (p < 0.05, n = 21, Figs. 2A and B). This finding suggests that PD-1 is not only expressed by exhausted T cells, but is also upregulated in T cells undergoing activation.

Figure 2.

Co-expression of BTLA and PD-1 by CD4⁺ T cells defines a population of exhausted T cells in HCC. (A and B) The capacity of BTLA^+/− and PD-1^+/− non-tumor and tumor CD4⁺ T cells subsets to produce IFNγ ex vivo (n = 13 for BTLA, n = 21 for PD-1). *p < 0.05 and **p < 0.01. (C and D) IFNγ-producing capacity of non-tumor and tumor-infiltrating CD4⁺ T cell subsets sub-classified by BTLA and PD-1 (n = 10). (E) Analysis of CCR7, CD45RA, CD25, and CTLA-4 expression on tumor-infiltrating CD4⁺ T cell subsets sub-classified by BTLA and PD-1 (n = 7). The dashed squares confine CD25^high or CTLA-4^high cells (E). Results are expressed as mean ± SEM and the p-values were calculated using the Student's t test (B and D).

Indeed, we observed that only BTLA⁺PD-1⁺ CD4⁺ T cells in tumors exhibited an exhausted phenotype with radically reduced IFNγ (Figs. 2C and D). In contrast, the BTLA⁻PD-1⁺ CD4⁺ T cells in tumors were the most potent in producing IFNγ among all CD4⁺ subsets (Fig. 2D). Of note, the intensity of PD-1 is much higher in BTLA⁺PD-1⁺ CD4⁺ T cells than that in BTLA⁻PD-1⁺ CD4⁺ T cells (Figs. 1C and 2C). These data suggest that BTLA⁺ can identify exhausted PD-1-expressing CD4⁺ T cell subset, whereas BTLA⁻ defines PD-1⁺ CD4⁺ T cells undergoing activation in human HCC. We also analyzed the phenotypes of BTLA⁺PD-1⁺ and BTLA⁻PD-1⁺ CD4⁺ T cells from HCC tissues. Both of these two PD-1-expressing subsets exhibited a CD45RA⁻CCR7⁻ effector memory phenotype with extensive expression of the CD25, whereas BTLA⁺PD-1⁺ subset have higher level of CTLA4 (Fig. 2E).

Distinct roles of PD-L1 blockade in HCC-derived BTLA⁺PD-1⁺ and BTLA⁻ PD-1⁺ CD4⁺ T cells

The results described above suggested that interaction of PD-1 and it physical ligand may result in distinct activated status in BTLA⁺PD-1⁺ and BTLA⁻ PD-1⁺ CD4⁺ T cells. To test this, we purified the BTLA⁺PD-1⁺ and BTLA⁻ PD-1⁺ CD4⁺ T cells from HCC tissues by sorting the cells directly ex vivo as described in Methods. We then cultured these cells alone or with tumor-derived autologous CD14⁺ cells (70% of which were PD-L1⁺)²⁵ in the presence or absence of an anti-PD-L1 blocking mAb. At the end of the co-culture, the ability of cytokines production was tested by short stimulation with Leukocyte Activation Cocktail as described in Methods.

Culturing alone ex vivo for 24 h partially restored the ability of BTLA⁺PD-1⁺ CD4⁺ T cells to produce IFNγ and TNF-α; however, BTLA⁻ PD-1⁺ CD4⁺ T cells cultured ex vivo did not produce increased level of IFNγ or TNF-α, but exhibited an attenuated production of those cytokines (Figs. 3A and B, left). This finding indicates that environmental factors of HCC is essential for maintaining the exhaustion of BTLA⁺PD-1⁺ CD4⁺ T cells and activation of BTLA⁻PD-1⁺ CD4⁺ T cells. Consistent with this, we found that adding of PD-L1-expressing tumor-derived CD14⁺ cells could effectively maintain the original IFNγ and TNF-α production profiles of BTLA⁺PD-1⁺ and BTLA⁻PD-1⁺ CD4⁺ T cells (Figs. 3A and B, right). Thus, PD-L1 signals may contribute to both early activation and subsequent exhaustion of CD4⁺ T cells in tumors. Accordingly, we demonstrated that blocking the PD-L1 signal in our culture system could partially reduce the exhaustion of BTLA⁺PD-1⁺ CD4⁺ T cells, as well as the production of IFNγ and TNF-α by BTLA⁻PD-1⁺ CD4⁺ T cells (Figs. 3A and B, right).

Figure 3.

Blockade of PD-L1/PD-1 pathway increases the frequency of cytokine-producing BTLA⁺PD-1⁺ CD4⁺ T cells. The BTLA⁺ PD-1⁺ and BTLA⁻ PD-1⁺ tumor CD4⁺ T cells were purified by FACS directly ex vivo prior to stimulation; autologous tumor monocytes (CD14⁺) were purified by MACS. (A) Sorted tumor CD4⁺ T cells subsets were analyzed by Flow cytometry directly ex vivo (first row on the left), or were cultured in medium for 24 h before evaluating intracellular cytokine production (second row on the left). FACS-purified BTLA⁺PD-1⁺ or BTLA⁻PD-1⁺ tumor CD4⁺ T cells were co-cultured with MACS-purified autologous tumor monocytes (CD14⁺) in the presence of anti-PD-L1 mAb or a control IgG (right). Intracellular IFNγ and TNF-α were analyzed by Flow cytometry (n = 3). The percentages of sorted tumor CD4⁺ T cells subsets expressing/co-expressing IFNγ and TNF-α was displayed in pie charts (B).

BTLA signals suppress CD4⁺ T cell function in HCC

After finding distinct effects of PD-1 signals in BTLA⁺PD-1⁺ and BTLA⁻PD-1⁺ CD4⁺ T cell function, we evaluated the influence of BTLA signals on tumor CD4⁺ T cell function. The ligand for BTLA was discovered to be HVEM.²⁶ We determined the expression of HVEM by immunohistochemical staining in paraffin-embedded HCC samples. In the samples analyzed, most cells, including hepatoma, hepatocytes and stroma cells, were positive for HVEM (Fig. 4A). To further confirm this finding, we subsequently used FACS to detect HVEM expression in hepatoma cell lines and tumor-infiltrating leukocytes. In support, all hepatoma cell lines we analyzed were positive for HVEM (Fig. 4B). Similar levels of HVEM expression were observed in tumor/non-tumor-infiltrating mononuclear cells, including CD14⁺ monocytes, CD3⁺ T cells, and CD19⁺ B cells (Fig. 4C). Therefore, the tumor cells as well as host immune cells could provide the sources of HVEM for the signaling of BTLA⁺ CD4⁺ T cells. Considering all tumor-infiltrating mononuclear cells could provide source of HVEM, we finally isolated tumor-infiltrating mononuclear cells directly ex vivo and cultured them in the presence of an anti-BTLA mAb or a control IgG. Treatment with anti-BTLA resulted in a higher percentage of cytokine-producing CD4⁺ T cells (Figs. 4D and E). Thus, BTLA is not only a marker for exhausted PD-1-expressing CD4⁺ T cells, but also a receptor that contribute to T cell suppression in HCC.

Figure 4.

Hepatoma and host immune cells express high levels of HVEM which suppresses CD4⁺ T cell function in HCC. (A) Paraffin-embedded hepatoma samples (n = 5) were stained for HVEM. Bar, 100 μm. (B) HVEM expression detected by FACS on hepatoma cell lines Hep3B, HepG2 and MHCC-97L. Cells were stained with isotype control (black line) or with anti-HVEM mAb (red line). (C) HVEM expression on monocytes, T cells and B cells derived from HCC tumors (left) or from both tumor and non-tumor tissues (right). The horizontal bars represent median values. (D and E) Tumor-infiltrating mononuclear cells were incubated overnight with anti-BTLA mAb or a control IgG before detection of intracellular cytokine production (n = 3). The percentages of CD4⁺ T cells expressing/co-expressing IFNγ and TNF-α was displayed in pie charts (E).

Materials and methods

Patients and specimens

49 patients with hepatocellular carcinomas underwent curative resection between 2008 and 2015 in the Third Affiliated Hospital of Sun Yat-sen University (Table 1), and samples from these patients were used for flow cytometry analysis, immunohistochemistry and functional assays. Paired fresh non-tumor (at least 3 cm away from the tumor site) and tumor tissues from these patients were used for the isolation of tissue-infiltrating leukocytes.²⁵ None of the patients had received anticancer therapy before the sampling, and individuals with concurrent autoimmune disease, HIV, or syphilis were excluded. Clinical stages were classified according to the guidelines of the International Union against Cancer. All samples were anonymously coded in accordance with local ethical guidelines (as stipulated by the Declaration of Helsinki), and written informed consent was obtained. The protocol was approved by the Review Board of the Third Affiliated Hospital of Sun Yat-sen University.

Isolation and sorting of mononuclear cells from tissues

Fresh non-tumor- and tumor-infiltrating mononuclear cells were obtained as previously described.²⁵ In short, tissue specimens were cut into small pieces and digested in RPMI1640 supplemented with 0.05% collagenase IV, 0.002% DNase I (Roche), and 20% FBS at 37°C for 20 min. Dissociated cells were filtered through a 150-μm mesh and separated by Ficoll density gradient centrifugation to obtain mononuclear cells. CD4⁺ T cells were enriched from the mononuclear cells by MACS suing a CD4 negative selection kits (Miltenyi Biotec) according to the manufacturer's instructions. For FACS sorting, CD4⁺-enriched cells were stained with anti-BTLA and anti-PD-1, and sorted on a BD Influx (Becton Dickinson). T cells were cultured in RPMI1640 supplied with 10 IU/mL IL-2 (eBioscience). Tumor CD14⁺ cells were purified from the mononuclear cells using anti-CD14 magnetic beads (Miltenyi Biotec) according to the manufacturer's instructions. Over 95% of tissue-derived CD14⁺ cells exhibited a CD14^hi phenotype.²⁵

Co-culture assay

The BTLA⁺ PD-1⁺ and BTLA⁻ PD-1⁺ tumor CD4⁺ T cells were purified directly ex vivo prior to stimulation, by MACS followed by FACS as described above; autologous tumor monocytes (CD14⁺) were purified by MACS. In one experiment, freshly isolated tumor CD4⁺ T cells subsets were analyzed directly ex vivo, or were cultured alone for 24 h before evaluating intracellular cytokine production (Fig. 3). Otherwise, FACS-purified BTLA⁺PD-1⁺ and BTLA⁻PD-1⁺ tumor CD4⁺ T cells were co-cultured with MACS-purified autologous tumor monocytes (CD14⁺) in the presence of a blocking mAb against PD-L1 (eBioscience) or a control IgG (R&D Systems) (Fig. 3). In some experiments, tumor-infiltrating mononuclear cells were pretreated with a blocking mAb against BTLA (eBiosciences) or a control IgG (R&D Systems) (Fig. 4).

Tumor cell lines cultures

Human hepatoma cell lines Hep3B and HepG2 were obtained from the ATCC and cultured according to the ATCC guidelines. Human hepatoma cell line MHCC-97L was obtained from Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). All cells were tested for mycoplasma contamination using a single-step PCR method.²⁵ They were maintained in DMEM supplemented with 10% FBS (HyClone), and used within 10 passages.

Flow cytometry

Human hepatoma cell lines were stained with fluorochrome-conjugated mAb for HVEM or control Ab (eBioscience). Nontumor- and tumor-infiltrating mononuclear cells and T cells were left untreated or stimulated at 37°C for 5 h with Leukocyte Activation Cocktail, with BD GolgiPlug™ (a ready-to-use polyclonal cell activation mixture containing the phorbol ester, PMA, Ionomycin and Brefeldin A. BD Pharmingen) according to the manufacturer's instructions. Thereafter, cells were stained with surface markers, fixed and permeabilized with IntraPrep reagents (Beckman Coulter), and finally stained with intracellular markers. Data were acquired on a Gallios Flow Cytometer (Beckman Coulter) and analyzed with FlowJo software. CD4⁺ T cells were gated based on CD3 and CD4⁺ expression. Other fluorochrome-conjugated Abs used are described in Table S1.

Immunohistochemistry

Paraffin-embedded samples were cut into 5-μm sections and processed for immunohistochemistry as previously described.²⁵ Following incubation with a control IgG or a mAb against human HVEM (R&D Systems), the sections were stained using the EnVision System with diaminobenzidine (DAKO). The tissue sections were screened in an inverted research microscope (model DM IRB; Leica) at low magnification (×100). Thereafter, the respective areas were measured at ×400 magnification.²⁷

Statistical analysis

Results are expressed as mean ± SEM unless otherwise indicated in the text. Statistical significance was determined by Student's t test. All data were analyzed using two-tailed tests unless otherwise specified, and a p value of less than 0.05 was considered statistically significant.

Discussion

PD-1 is a hallmark of both exhausted and activated T cells.^1,2,28 In this study, we observe that PD-1⁺ CD4⁺ T cells co-expressing BTLA not only represent the most abundant tumor-infiltrating CD4⁺ T cells in patients with HCC but also represent the most dysfunctional population. BTLA⁺ PD-1⁺ tumor-infiltrating CD4⁺ T cells are selectively correlated with advanced stage HCC. Strikingly, PD-L1 blockade effective restores the function of BTLA⁺PD-1⁺ tumor CD4⁺ T cells but partially suppresses the activation of BTLA⁻PD-1⁺ CD4⁺ T cells. These findings suggest that BTLA can be introduced to sub-classify whether the PD-1-expressing tumor T cells are “exhausted” or “activated.”

Tumor progression is now recognized as the product of evolving cross-talk between different cell types within the tumor and its stroma.^29-32 Although normal stroma is nonpermissive for neoplastic progression, cancer cells can modulate adjacent stroma to generate a supportive microenvironment.^33-37 This includes the ability to alter the ratios of effector to regulatory T cells and to affect the functions of APCs and the expression of co-signaling molecules, which in turn creates an immunosuppressive network to promote tumor progression and immune evasion.^29,30,38 In the current study, we observed marked upregulation of both co-inhibitory receptors PD-1 and BTLA in human tumor CD4⁺ T cells. Frequency of tumor BTLA⁺PD-1⁺ CD4⁺ T cells was significantly increased in patients with advanced stage HCC, which agrees with the general view that co-inhibitory receptor PD-1 facilitates the development of cancer. However, we also found that the PD-1⁺BTLA⁻ CD4⁺ T cells in tumor tissues were the most potent in producing IFNγ, which suggests that PD-1 is not only expressed by exhausted T cells, but also upregulated in T cells undergoing activation.

PD-1 has been the primary marker for exhausted T cells.^1,28,39,40 However, emerging evidence show that PD-1 expressing tumor-infiltrating T cells is indeed a favorable prognostic biomarker in some cancers.^12-14 Our data suggest that PD-1 is an imperfect marker of CD4⁺ T cell exhaustion and that co-expression of BTLA and PD-1 clearly marks the T cells with the most exhausted phenotype. Several of our observations support this notion. First, BTLA⁺PD-1⁺ tumor CD4⁺ T cells were more dysfunctional than BTLA⁻PD-1⁺ and BTLA⁻PD-1⁻ tumor CD4⁺ T cells, as they could produce significantly less IFNγ and TNF-α ex vivo. Second, although all PD-1-expressing CD4⁺ T cells exhibited a CD45RA⁻CCR7⁻ effector memory phenotype with extensive expression of CD25, BTLA⁺PD-1⁺ CD4⁺ T cells have significantly higher levels of CTLA-4. Third, blockade of PD-1/PD-L1 interaction in tumor BTLA⁺PD-1⁺ CD4⁺ T cells significantly increased the capacity of cells to produce IFNγ/TNF-α, whereas such treatment did attenuate the IFNγ/TNF-α production in BTLA⁻PD-1⁺ CD4⁺ T cells. Fourth, unlike PD-1⁺CD4⁺ T cells from non-tumor liver, only PD-1⁺CD4⁺ T cells from tumor tissue co-express high level of BTLA in patients with advanced HCC. Thus, the exhaustion of CD4⁺ T cells is a process of complexity and may be regulated by multiple inhibitory receptors.^41,42 This hypothesis is compatible with previous studies showing that simultaneous upregulation of Tim-3 and PD-1 is associated with tumor antigen-specific CD8⁺ T cell dysfunction in melanoma of human and mice.^24,3-46

It has been established that BTLA is a lymphocyte inhibitory receptor with similarities to CTLA-4 and PD-1.^19,47 However, unlike PD-1 and CTLA-4, BTLA displays T cell inhibition via interaction with tumor necrosis family receptors. Tumor necrosis factor (receptor) superfamily, member 14 (TNFRSF14), also known as HVEM, negatively regulate BTLA-expressing T cell function.^26,48 The extensive and broad expression of HVEM on tumor cells as well as host immune cells in HCC increases the likelihood of engagement of this pathway, that is, the probability that BTLA⁺ CD4⁺ T cells will encounter HVEM⁺ cells in HCC environments. As a result, the dysfunction of T cells in tumors helps avoid immune response against tumors. Therefore, our results show that BTLA is not only a marker that identify exhausted PD-1-expessing CD4⁺ T cells, but also functionally suppress CD4⁺ T cell function through interaction with HVEM⁺ cells. At present, the regulating mechanisms of PD-1 and BTLA expression in T cells in human cancer are still unclear.^49,50 Both BTLA⁻PD-1⁺ and BTLA⁺PD-1⁺ CD4⁺ T cells have extensive expressions of CTLA-4 and CD25. Accordingly, the prolonged and chronic exposure of T cells to activated signals during tumor progression also results in PD-1- and BTLA-mediated T cell dysfunction. Studying the mechanism involved in T cell activation may reveal the conditions that lead to T cell dysfunction. In other words, if the factors that lead to PD-1 and BTLA upregulation in tumors are abolished, the conditions for triggering antitumor T cell responses also significantly impaired. Therefore, it is very important for us to sub-classify whether T cell responses have become “exhausted” or are still “activated.”^15,16

In summary, our data confirm the dual role of PD-1 as a marker of both exhausted and activated T cells in human cancers, and here in HCC. We demonstrate that BTLA⁺ can identify exhausted PD-1-expressing CD4⁺ T cell subset, whereas BTLA⁻ defines PD-1⁺ CD4⁺ T cells undergoing activation in human HCC. Immunotherapy of cancer by blocking the interaction between PD-1 and PD-L1 might be selectively applied to patients with significant infiltration of BTLA⁺PD-1⁺ T cells.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This study was supported by the Pearl River S&T Nova Program of Guangzhou (201605122020199), the Natural Science Fund of Guangdong Province (2014A030313089), the NSFC (81422036, 81503317, 81202319 and 81672701), Guangdong Province Higher Vocational Colleges & Schools Pearl River Scholar Funded Scheme (2016), and the Medical Scientific Research Foundation of Guangdong Province (A2014237).